Immersive Experience of Movie Scenes Based on Convolutional Neural Network

Abstract

This paper discusses how to use virtual reality (VR) and computer aided design (CAD) technology to enhance the immersive experience of movie scenes. Firstly, DL technology, especially the model of Convolutional Neural Network (CNN), is used to classify and identify the interactive objects in the virtual movie scene. Then, the deep reinforcement learning algorithm is used to simulate and experiment these objects to determine the best interaction mode. Finally, the best interactive effect is generated by combining the input information of the audience and the results of the deep reinforcement learning algorithm. CNN is obviously superior to the 3D modeling algorithm based on wavelet neural network (WNN) model in accuracy and recall. This improvement shows that CNN can more accurately identify the key elements and details in VR movie scenes, and has better scene layout optimization ability. The focus of the research is to create a highly realistic virtual movie scene by combining deep learning (DL) technology and computer graphics, so that the audience can interact with the VR scene more naturally. This interactive design can improve the immersive experience of the movie scene and enhance the audience's viewing experience.